Digestive tract early cancer onset risk early warning method and system based on data analysis

Abstract

The invention belongs to the technical field of medical health monitoring, and discloses a digestive tract early cancer onset risk early warning method and system based on data analysis. Comprising the steps that a user side collects user feature data and sends the user feature data to a cloud platform; the cloud platform analyzes the user feature data and judges the digestive tract state of the user; the cloud platform estimates the digestive tract early cancer probability according to the digestive tract state and sends the digestive tract early cancer probability to the medical terminal; the medical terminal collects user inspection data and sends the user inspection data to the cloud platform; the cloud platform evaluates the early cancer degree of the digestive tract in combination with the user examination data and the user feature data; the cloud platform generates rehabilitation suggestions according to the early cancer degree of the digestive tract and sends the rehabilitation suggestions to the corresponding user side; the digestive tract cancer can be found and diagnosed as soon as possible, and treatment or protection measures can be taken in the early cancer stage of the digestive tract, so that the possibility of disease development or complications is reduced, and the rehabilitation effect and the long-term health condition of a user are improved.

Description

Technical Field

[0001] The present invention relates to the field of medical health monitoring technology, and more specifically, to a method and system for early warning of the risk of early gastrointestinal cancer based on data analysis. Background Art

[0002] Digestive tract diseases are a serious threat to human health with high morbidity and mortality rates. Therefore, early detection and diagnosis are crucial to improving patients' survival rates. However, since there are often no obvious symptoms in the early stages of digestive tract diseases or the symptoms are similar to those of other common diseases, patients are not aware of them, which delays medical treatment. Therefore, there is an urgent need to develop an effective early warning method to help doctors and patients detect and diagnose early digestive tract cancer in a timely manner.

[0003] The patent with announcement number CN118522467B discloses a method and system for analyzing digestive tract health data; including: based on digestive tract endoscopy image data, denoising the image through a Gaussian filter, and using histogram equalization to enhance image contrast, applying a convolutional neural network, and extracting visual features from the image through a combination of convolutional layer, ReLU activation layer and pooling layer to generate visual feature data; this invention combines convolutional neural network with Faster R-CNN to improve the accuracy and speed of lesion area detection, achieve fast and accurate identification, long short-term memory network analysis of time series data, predict disease progression, support early diagnosis, random forest algorithm to identify abnormal patterns, enhance disease warning, topological data analysis to reveal lesion structure and association, and decision tree algorithm combined with medical knowledge graph to provide personalized treatment recommendations, considering comprehensive benefits and risks, to achieve more accurate and personalized diagnosis and treatment.

[0004] The above technology realizes digestive disease analysis based on digestive tract endoscopy data; however, digestive tract endoscopy data needs to be obtained by professional doctors with rich medical experience through endoscopic examination, the examination cost is high, and the endoscopic examination process may cause discomfort to patients; therefore, patients develop a resistance to endoscopic examinations, causing patients to postpone or avoid regular examinations without feeling any discomfort, thereby delaying the early detection and diagnosis of digestive tract cancer, and ultimately posing a serious risk to the patient's life safety.

[0005] In view of this, the present invention proposes a method and system for early gastrointestinal cancer risk warning based on data analysis to solve the above problems. Summary of the invention

[0006] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for early warning of the risk of early gastrointestinal cancer based on data analysis, applied in a cloud platform, comprising:

[0007] Collect user characteristic data;

[0008] Analyze user characteristic data to determine the user's digestive tract status;

[0009] Estimate the probability of early digestive tract cancer based on digestive tract status;

[0010] Collect user inspection data;

[0011] Combine user examination data and user characteristic data to assess the degree of early digestive tract cancer;

[0012] Generate rehabilitation recommendations based on the degree of early digestive tract cancer.

[0013] Furthermore, the digestive tract status includes normal, slightly abnormal, moderately abnormal and severely abnormal; if the digestive tract status is slightly abnormal, moderately abnormal or severely abnormal, the probability of early digestive tract cancer is estimated; if the digestive tract status is normal, the probability of early digestive tract cancer is not estimated; the probability of early digestive tract cancer is the probability corresponding to each type of digestive tract cancer;

[0014] The user characteristic data includes physiological data, life data, stress data and metabolic data; marking the non-numerical data in the user characteristic data as text data, setting different digital labels for different text data and marking them as data labels; replacing the text data in the user characteristic data with the corresponding data labels, and marking the replaced user characteristic data as replacement characteristic data;

[0015] The steps to estimate the probability of early digestive tract cancer include:

[0016] Step a1: Establish fuzzy sets, and divide each data in the replacement feature data into multiple fuzzy sets;

[0017] Step a2: convert each data in the replacement feature data into the membership degree of each corresponding fuzzy set through fuzzification technology; fuzzification is the process of converting an exact numerical value into the membership degree corresponding to the fuzzy set;

[0018] Step a3: define fuzzy rules;

[0019] Step a4: matching the fuzzified replacement feature data with the fuzzy rules, performing fuzzy reasoning, and obtaining fuzzy reasoning results, where the fuzzy reasoning results are the membership degrees corresponding to each type of digestive tract cancer;

[0020] Step a5: Use the fuzzy reasoning result as the probability of early digestive tract cancer.

[0021] Furthermore, the method for determining the digestive tract status of the user includes:

[0022] Input the replacement feature data into the trained state judgment model to predict the corresponding state label, where the state label is a digital label corresponding to the digestive tract state; and judge the user's digestive tract state based on the predicted state label;

[0023] The training process of the state judgment model includes:

[0024] Collect b groups of replacement feature data in advance, set corresponding state labels for the b groups of replacement feature data, where b is an integer greater than 1, and convert the replacement feature data and the corresponding state labels into a corresponding set of feature vectors;

[0025] Each group of feature vectors is used as the input of the state judgment model. The state judgment model takes a group of predicted state labels corresponding to each group of replacement feature data as output, and takes the actual state labels corresponding to each group of replacement feature data as prediction targets, where the actual state labels are pre-set state labels corresponding to the replacement feature data; minimizing the sum of prediction errors of all replacement feature data is used as the training target; the state judgment model is trained until the sum of prediction errors reaches convergence and the training is stopped; the state judgment model is a deep neural network model.

[0026] Furthermore, a probability threshold is preset, and the probability corresponding to each digestive tract cancer is compared with the probability threshold, and cancers with a probability greater than or equal to the probability threshold are marked as cancers to be detected, and cancers with a probability less than the probability threshold are not marked; the degree of early digestive tract cancer is the risk degree corresponding to each cancer to be detected;

[0027] The steps of evaluating the extent of early digestive tract cancer include:

[0028] Step d1: preset the initial search interval [a0, b0]; set the segmentation ratio h;

[0029] Step d2: According to the segmentation ratio h, the initial search interval [a0, b0] is divided into two new points ζ1 and ζ2, ζ1 = a0 + (1-h)(b0-a0), ζ2 = a0 + h(b0-a0); different digital labels are set for different cancers to be detected and marked as cancer labels; user inspection data and replacement feature data are marked as analysis data;

[0030] Step d3: Randomly select a cancer label that is not marked as a selected label and mark it as the evaluation label;

[0031] Step d4: respectively obtain the matching degrees corresponding to ζ1 and ζ2, mark the matching degree corresponding to ζ1 as the first matching degree, and mark the matching degree corresponding to ζ2 as the second matching degree;

[0032] Step d5: If the first matching degree is greater than or equal to the second matching degree, the minimum value of the updated search interval is the minimum value of the search interval before the update, and the maximum value is the point corresponding to the second matching degree; if the first matching degree is less than the second matching degree, the minimum value of the updated search interval is the point corresponding to the first matching degree, and the maximum value is the maximum value of the search interval before the update;

[0033] Step d6: Determine whether to return to step d4; if so, return to step d4, if not, proceed to step d7;

[0034] Step d7: Determine whether all cancer tags are marked as selected tags. If so, obtain the degree of early digestive tract cancer; if not, return to step d3.

[0035] Furthermore, in step d4, the method for respectively obtaining the matching degree corresponding to ζ1 and ζ2 is as follows: ζ1, analysis data and evaluation label are used as the first test data, and ζ2, analysis data and evaluation label are used as the second test data; the first test data and the second test data are respectively input into the trained matching analysis model to predict the corresponding matching degree; the matching degree is the matching degree between the risk degree of the cancer to be detected corresponding to the analysis data and the evaluation label; the training process of the matching analysis model is consistent with the training process of the state judgment model, and both are deep neural network models;

[0036] In step d6, the method for determining whether to return to step d4 is: presetting a width threshold, subtracting the minimum value from the maximum value in the updated search interval to obtain the interval width; comparing the interval width with the width threshold; if the interval width is greater than or equal to the width threshold, then re-dividing the updated coefficient search area into two new points ζ1 and ζ2 according to the segmentation ratio h, and returning to step d4; if the interval width is less than the width threshold, then taking the average of the maximum and minimum values ​​in the updated coefficient search area as the evaluation label corresponding to the risk level of the cancer to be detected; and marking the evaluation label as the selected label, and entering step d7.

[0037] Furthermore, an evolutionary algorithm is used to generate rehabilitation suggestions, the steps of which include:

[0038] Step p1: Collect historical rehabilitation suggestions, set different digital labels for different rehabilitation suggestions in the historical rehabilitation suggestions, and mark them as suggestion labels;

[0039] Step p2: randomly generate an initial antibody group, where the antibodies in the initial antibody group correspond one-to-one to the recommended labels;

[0040] Step p3: determine the fitness function;

[0041] Step p4: Calculate the fitness of each antibody in the initial antibody group, and perform antibody screening to generate a memory antibody group;

[0042] Step p5: Calculate the corresponding clone size for each antibody in the memory antibody group, and clone it to generate a clone antibody group;

[0043] Step p6: Calculate the selection probability of each antibody in the cloned antibody group, and determine whether to select the antibody based on the selection probability to generate a selected antibody group;

[0044] Step p7: using an interpolation method to generate new antibodies for the antibodies in the selected antibody group to generate an interpolated antibody group;

[0045] Step p8: performing antibody recombination on the antibodies in the selected antibody group to generate a recombinant antibody group;

[0046] Step p9: merging the interpolated antibody group and the recombinant antibody group to generate a merged antibody group;

[0047] Step p10: clustering and screening the antibodies in the combined antibody group to generate a clustered antibody group;

[0048] Step p11: Determine whether the iteration is finished; if so, calculate the fitness of each antibody in the clustered antibody group, and obtain the rehabilitation advice corresponding to the recommended label corresponding to the antibody with the largest fitness; if not, regenerate the memory antibody group based on the clustered antibody group, and return to step p5.

[0049] Furthermore, in step p2, the initial antibody group X(0) = {X1, X2, X3, ... X q}, X q is the qth antibody, that is, the size of the initial antibody population is q; the range of antibodies is historical recovery suggestions; the number of iterations t of the initial antibody population is 0;

[0050] In step p3, the fitness function is expressed as: f=kx; wherein f is fitness, and kx is rehabilitation effect; the method for obtaining the rehabilitation effect is: taking the recommended label and analysis data corresponding to the antibody as evaluation data, inputting the evaluation data into the trained effect prediction model, and predicting the corresponding rehabilitation effect, the training process of the effect prediction model is consistent with the training process of the state judgment model, and both are deep neural network models; the rehabilitation effect is the degree of decrease in the degree of early digestive tract cancer after the user executes the rehabilitation suggestion;

[0051] In step p4, the method of antibody screening is: screening a antibodies with the largest fitness from the initial antibody group, q>a>1;

[0052] In step p5, the method for calculating the clone scale includes:

[0053]

[0054] Φ i =min(exp(||X i -X j ||));

[0055] In the formula, R i is the clone size of the i-th antibody, Int is the upward rounding function, q is the initial antibody population size, Φ i is the affinity of the ith antibody, f i is the fitness of the ith antibody, min is the minimum function, exp is the natural exponential function, ||X i -X j || is the Euclidean distance between the i-th antibody and the j-th antibody, i≠j, j∈[1,q].

[0056] Furthermore, in step p6, the method for generating the selected antibody group includes:

[0057] Calculate the fitness of each antibody in the cloned antibody group, and add them up in sequence to obtain the total antibody fitness, divide the fitness of each antibody by the total antibody fitness, and obtain the selection probability of each antibody; sort the selection probability of each antibody from large to small, and replace each selection probability in order according to the positive order with the sum of each selection probability and all selection probabilities before the corresponding selection probability; obtain all selection probabilities after replacement and mark them as replacement probabilities; the antibody corresponding to the replacement probability is consistent with the antibody corresponding to the selection probability before replacement; regard every two adjacent replacement probabilities as an analysis set according to the positive order; mark the replacement probability with the largest value in each analysis set as the first probability , mark the replacement probability with a smaller value as the second probability; construct the probability screening range corresponding to each analysis set according to the first probability and the second probability of each analysis set, wherein the first probability is the maximum value of the probability screening range, the second probability is the minimum value of the probability screening range, and the minimum value is an open interval, and the maximum value is a closed interval; each analysis set corresponds to the antibody corresponding to the second probability; wherein the minimum value of the probability screening range corresponding to the first antibody is 0, and the maximum value is the corresponding selection probability; randomly generate a random number between (0, 1], select the antibody corresponding to the probability screening range where the a random number is located, and generate a selected antibody group;

[0058] In step p7, the method for generating an interpolated antibody group includes:

[0059] Calculate the fitness of each antibody in the selected antibody group, and sort them from large to small to generate a sorting table; select half of the antibodies in the selected antibody group according to the positive order of the sorting table, and mark them as screening antibodies; use the interpolation method to generate new antibodies for every two adjacent screening antibodies; calculate the fitness corresponding to each new antibody, and add it to the sorting table, retain the antibodies corresponding to the first y fitnesses in the sorting table, and generate an interpolated antibody group, where y is the number of antibodies in the selected antibody group;

[0060] The expression of the new antibody is: X X =(1-α′)X′+α′X″; where X X is a new antibody, X′ and X″ are two adjacent screened antibodies, α′ is the interpolation coefficient, and α′ is a random number between [0, 1].

[0061] Furthermore, in step p8, the method for antibody recombination comprises:

[0062] Randomly select Y antibodies from the selected antibody group as parent antibodies for antibody recombination to generate progeny antibodies S c , until the number of unselected antibodies in the selected antibody group is less than Y, the antibody recombination is completed; the recombinant antibody group includes multiple progeny antibodies S c ;

[0063] In step p10, the method for generating a clustered antibody group includes: clustering the antibodies in the merged antibody group using a clustering algorithm to obtain u clusters, where u is an integer greater than 1; adding the antibodies in each cluster in turn, and then dividing by the number of antibodies in the corresponding cluster to obtain the mean corresponding to each cluster, and marking it as the cluster mean; subtracting the corresponding cluster mean from the antibodies in each cluster in turn to obtain the antibody difference corresponding to each antibody; comparing the antibody differences corresponding to the same cluster, obtaining the antibody difference with the minimum value corresponding to each cluster, and marking it as the minimum difference; taking the antibody corresponding to the minimum difference corresponding to each cluster as the center of the corresponding cluster; generating a clustered antibody group according to the centers corresponding to the u clusters;

[0064] In step p11, the method for determining whether the iteration is finished includes:

[0065] An iteration threshold is preset, and the number of iterations t is compared with the iteration threshold; if the number of iterations t is greater than or equal to the iteration threshold, the iteration ends; if the number of iterations t is less than the iteration threshold, the maximum fitness in the combined antibody group is marked as the single maximum fitness; the single maximum fitness in the current iteration process is subtracted from the single maximum fitness in the previous iteration process and the absolute value is taken as the fitness change;

[0066] Preset variable threshold and quantity threshold; compare the fitness change with the variable threshold, if the fitness change is greater than or equal to the variable threshold, do not generate a smooth instruction; if the fitness change is less than the variable threshold, generate a smooth instruction; count the number of smooth instructions generated and compare with the quantity threshold; if the number of smooth instructions is less than the quantity threshold, the iteration continues, and the number of iterations t=t+1, if u<a, randomly generate au new antibodies and form a new memory antibody group with the clustered antibody group, if u≥a, select a antibodies with the largest fitness from the clustered antibody group to form a new memory antibody group; if the number of smooth instructions is greater than or equal to the quantity threshold, the iteration ends.

[0067] A digestive tract early cancer risk warning system based on data analysis, implementing the digestive tract early cancer risk warning method based on data analysis, applied to a cloud platform, including:

[0068] A data collection module, used to collect user characteristic data;

[0069] The data analysis module is used to analyze the user's characteristic data and determine the user's digestive tract status;

[0070] The probability estimation module estimates the probability of early digestive tract cancer based on the digestive tract status;

[0071] A data acquisition module, used to collect user inspection data;

[0072] The risk assessment module is used to combine user examination data and user characteristic data to assess the degree of early digestive tract cancer;

[0073] The suggestion generation module generates rehabilitation suggestions based on the degree of early digestive tract cancer.

[0074] The technical effects and advantages of the method and system for early warning of the risk of early gastrointestinal cancer based on data analysis of the present invention are as follows:

[0075] By collecting user feature data and using deep learning technology to accurately judge the state of the digestive tract, the user's potential digestive tract diseases can be identified in a timely manner, prompting the user to conduct corresponding examinations in a timely manner; fuzzy logic control methods and search algorithms are used to accurately estimate the probability and risk of early digestive tract cancer, detect and diagnose digestive tract cancer early, and effectively avoid life safety risks caused by delayed treatment; at the same time, evolutionary algorithms are used to adaptively generate rehabilitation suggestions, optimize health management effects, and improve the screening and monitoring efficiency of high-risk groups, which is helpful to take treatment or protective measures in the early stage of digestive tract cancer, thereby improving treatment effects, reducing the possibility of disease progression or complications, and thus improving the user's rehabilitation effects and long-term health status. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 This is a schematic diagram of a digestive tract early cancer risk warning system based on data analysis according to Example 1 of the present invention;

[0077] Figure 2 This is a flow chart of a method for evaluating the degree of early digestive tract cancer according to Example 1 of the present invention;

[0078] Figure 3 This is a flow chart of a method for generating rehabilitation suggestions according to Embodiment 1 of the present invention;

[0079] Figure 4 This is a flow chart of a method for early warning of the risk of gastrointestinal cancer based on data analysis according to Example 2 of the present invention. DETAILED DESCRIPTION

[0080] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0081] Example 1

[0082] See also Figure 1 As shown, the early digestive tract cancer risk warning system based on data analysis described in this embodiment includes a data acquisition module, a data analysis module, a probability estimation module, a data acquisition module, a risk assessment module and a suggestion generation module; each module is connected by wired and / or wireless means to achieve data transmission between modules;

[0083] The data collection module is used to collect user feature data on the user side and send it to the cloud platform.

[0084] The cloud platform is a service and application platform based on cloud computing technology, with a one-to-one correspondence between the user end and the user.

[0085] User characteristic data includes physiological data, life data, stress data and metabolic data; user characteristic data is relevant data that describes the user's health status and life habits.

[0086] Physiological data such as heart rate, blood pressure, blood oxygen level, etc.; digestive system discomfort or inflammation may affect the autonomic nervous system and cause abnormal heart rate; hypertension is related to metabolic syndrome, obesity, etc., which are risk factors for digestive tract cancers such as gastric cancer; fluctuations in blood oxygen levels usually reflect the oxygenation status of the user's body. Chronic inflammation, anemia or other chronic diseases will cause a decrease in blood oxygen levels, indirectly indicating the risk of digestive tract diseases; physiological data is obtained by users wearing wearable devices, such as smart watches, health bracelets, etc.

[0087] Life data includes dietary data, behavioral data and medication data; dietary data includes food types, frequency of meals, etc.; dietary data is one of the main risk factors for digestive tract cancer. Long-term consumption of high-salt, pickled, and smoked foods, as well as lack of vegetables and fruits, will increase the risk of gastric cancer and esophageal cancer; irregular diet, especially overeating or eating too little, will lead to disordered gastric acid secretion, increase the burden on the gastrointestinal tract, and thus increase the risk of digestive tract diseases; behavioral data includes smoking frequency, drinking frequency, etc.; smoking is an important risk factor for esophageal cancer and gastric cancer. Harmful substances in tobacco directly damage the mucosa of the esophagus and stomach, increasing the risk of cancer; long-term drinking, especially high-concentration liquor, will damage the gastric mucosa and esophageal mucosa, leading to inflammation and increased risk of cancer; medication data includes drug types, frequency of use, etc.; long-term use of some drugs (such as non-steroidal anti-inflammatory drugs, proton pump inhibitors, etc.) will increase the occurrence of gastric ulcers or gastritis, and long-term gastric mucosal damage is directly related to the increased risk of gastric cancer; life data is obtained by manual records by users.

[0088] Stress data is the user's stress level; long-term high stress will affect digestive function, leading to increased gastric acid secretion, causing problems such as gastric ulcers and gastritis, and even leading to early lesions of gastric cancer; stress will also affect the user's behavioral habits, such as smoking, drinking, overeating, etc., further increasing the risk of digestive tract diseases; stress data is obtained by users wearing wearable devices; wearable devices analyze stress levels by measuring the user's heart rate variability (HRV), and combine the user's sleep and exercise data to determine the long-term stress state.

[0089] Metabolic data such as blood sugar levels, lipid data, etc.; lipid data such as total cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), triglycerides, etc.; hyperglycemia and diabetes are related to gastrointestinal cancers such as gastric cancer and esophageal cancer. Long-term high blood sugar status in diabetic patients will lead to increased inflammation in the body, thereby increasing the risk of cancer; lipid metabolism disorders, that is, abnormal indicators such as total cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), triglycerides, especially hyperlipidemia and obesity, are risk factors for gastrointestinal diseases such as gastric cancer; blood sugar levels are obtained through smart blood glucose meters, and lipid metabolism is obtained through portable blood lipid testers.

[0090] The data analysis module is used on the cloud platform to analyze user characteristic data and determine the user's digestive tract status.

[0091] Methods for determining the digestive tract status of a user include:

[0092] Mark the non-numeric data in the user feature data as text data, set different digital labels for different text data, and mark them as data labels; text data such as food types, medicine types, etc.; replace the text data in the user feature data with the corresponding data labels, and mark the replaced user feature data as replacement feature data; input the replaced feature data into the trained state judgment model to predict the corresponding state label, which is a digital label corresponding to the digestive tract state, and the digestive tract state includes normal, slightly abnormal, moderately abnormal and severely abnormal; illustratively, the state label corresponding to normal is 1, the state label corresponding to slightly abnormal is 2, the state label corresponding to moderate abnormal is 3, and the state label corresponding to severely abnormal is 4; judge the user's digestive tract state according to the predicted state label.

[0093] The training process of the state judgment model includes:

[0094] Collect b groups of replacement feature data in advance, set corresponding status labels for the b groups of replacement feature data, where b is an integer greater than 1, and convert the replacement feature data and the corresponding status labels into a corresponding set of feature vectors; the status labels corresponding to the replacement feature data are collected by technicians in the historical digestive tract early cancer risk analysis process, and each group of replacement feature data is analyzed in turn based on actual experience to determine the corresponding digestive tract state, and the corresponding status labels are set for the b groups of replacement feature data in turn;

[0095] Each set of feature vectors is used as the input of the state judgment model. The state judgment model uses a set of predicted state labels corresponding to each set of replacement feature data as output, and uses the actual state labels corresponding to each set of replacement feature data as the prediction target. The actual state labels are the pre-set state labels corresponding to the replacement feature data. The training goal is to minimize the sum of the prediction errors of all replacement feature data. The calculation formula of the prediction error is η K =(β K -ε K ) 2 , where η K is the prediction error, K is the group number of the feature vector corresponding to the replaced feature data, β K Replace the predicted state label corresponding to the Kth group of feature data, ε K The actual state label corresponding to the feature data of the Kth group is replaced; the state judgment model is trained until the sum of the prediction errors reaches convergence and the training is stopped.

[0096] The above-mentioned state judgment model is specifically a deep neural network model; it includes an input layer, a hidden layer and an output layer; each hidden layer includes multiple neurons, each neuron is connected to the neurons in the next layer, and the connection contains weights, which determine the importance and influence of data transmission in the neural network; an activation function is applied to each neuron between the hidden layer and the output layer, and the activation function introduces nonlinearity, allowing the network to learn more complex patterns and features.

[0097] It should be understood that the purpose of judging the user's digestive tract condition is to promptly identify the user's potential digestive tract diseases, thereby reminding the user to pay attention to their own health in a timely manner, promoting the user to actively seek medical treatment, improving the user's health awareness, and achieving better health management and disease prevention effects.

[0098] The probability estimation module is used on the cloud platform to estimate the probability of early digestive tract cancer based on the digestive tract status and send it to the medical end.

[0099] If the digestive tract status is slightly abnormal, moderately abnormal or severely abnormal, the probability of early digestive tract cancer is estimated; if the digestive tract status is normal, the probability of early digestive tract cancer is not estimated; the probability of early digestive tract cancer is the probability corresponding to each type of digestive tract cancer; digestive tract cancers include gastric cancer, esophageal cancer, colon cancer, rectal cancer, etc.; the medical end refers to medical institutions, and medical institutions correspond to the medical end one by one.

[0100] The steps to estimate the probability of early digestive tract cancer include:

[0101] Step a1: Establish fuzzy sets, and divide each data in the replacement feature data into multiple fuzzy sets; for example, the fuzzy sets corresponding to heart rate are low heart rate, medium heart rate, high heart rate, etc., the fuzzy sets corresponding to food types are high health, medium health, low health, etc., and the fuzzy sets corresponding to the number of meals are few times, medium times, many times, etc.;

[0102] Step a2: convert each data in the replacement feature data into the membership of each corresponding fuzzy set through fuzzification technology; fuzzification is the process of converting an exact value into the membership corresponding to a fuzzy set, and fuzzification technologies include triangular membership function, trapezoidal membership function, etc.; for example, the value of the heart rate is low, and it is inferred that the membership of a low heart rate is 0.9, the membership of a medium heart rate is 0.1, and the membership of a high heart rate is 0;

[0103] Step a3: define fuzzy rules. The fuzzy rules are defined based on expert knowledge or relevant literature. For example, if the heart rate is low, the health is low, and the number of times is high, it is inferred that the risk of gastric cancer is high; if the heart rate is medium, the health is low, and the number of times is low, it is inferred that the risk of esophageal cancer is high.

[0104] Step a4: Match the fuzzified replacement feature data with the fuzzy rules, perform fuzzy reasoning, and obtain the fuzzy reasoning result, which is the membership degree corresponding to each cancer in the digestive tract cancer; the fuzzy reasoning method is, for example, the Mamdani or Sugeno fuzzy reasoning method; the fuzzy reasoning result is, for example, the membership degree of esophageal cancer is 0.1, the membership degree of colon cancer is 0.3, the membership degree of gastric cancer is 0.6, and the membership degree of rectal cancer is 0;

[0105] Step a5: Use the fuzzy reasoning result as the probability of early digestive tract cancer.

[0106] It should be noted that the reason for using the fuzzy logic control method to estimate the probability of early gastrointestinal cancer is that the fuzzy logic control method can convert precise replacement feature data into membership in a fuzzy set, thereby effectively handling uncertainty and ambiguity, allowing the system to more flexibly respond to complex early gastrointestinal cancer risk warning processes; at the same time, the fuzzy logic control method combines fuzzy rules and reasoning methods, allowing the system to make intelligent decisions based on expert knowledge, and can independently reason and make decisions based on the replacement feature data obtained in real time, reducing human intervention, thereby improving the automation and intelligence level of the production process; in addition, the fuzzy logic control method can be adjusted according to new data and knowledge, and flexibly modify rules and membership functions; enabling the system to adapt to changing medical knowledge and patient conditions, and enhancing applicability and scalability.

[0107] It should be understood that the purpose of estimating the probability of early gastrointestinal cancer is to facilitate medical institutions to conduct targeted medical examinations on users based on the probability of early gastrointestinal cancer, thereby improving the efficiency and accuracy of cancer screening, optimizing medical resources, and avoiding unnecessary examinations on users.

[0108] The data acquisition module is used by the medical side to collect user examination data and send it to the cloud platform.

[0109] The user examination data is the data obtained after the user undergoes various medical examinations in a medical institution; a probability threshold is preset, and the probability threshold is preset by a technician in the field according to the actual situation; the probability corresponding to each type of digestive tract cancer is compared with the probability threshold, and the cancer with a probability greater than or equal to the probability threshold is marked as a cancer to be detected, and the cancer with a probability less than the probability threshold is not marked; the medical institution performs corresponding medical examinations on the user according to the cancer to be detected, and obtains the corresponding user examination data; user examination data such as tumor markers (such as carcinoembryonic antigen, alpha-fetoprotein, etc.), alanine aminotransferase, aspartate aminotransferase, Red blood cell count, etc.; carcinoembryonic antigen is a glycoprotein, often used to monitor the risks of colon cancer, gastric cancer, pancreatic cancer, etc.; alpha-fetoprotein is mainly used for screening and monitoring of liver cancer, especially in high-risk groups (such as patients with chronic hepatitis); alanine aminotransferase is a liver-specific enzyme, and aspartate aminotransferase is an enzyme present in multiple tissues such as the liver and heart, both of which reflect the health of the liver. The ratio of the two is used to assess the nature of liver lesions (such as alcoholic liver disease or non-alcoholic fatty liver disease); the red blood cell count reflects the oxygen transport capacity of the blood. A low level may indicate anemia, which is common in chronic diseases, bleeding caused by tumors, or malnutrition.

[0110] The risk assessment module is used on the cloud platform to combine user examination data and user characteristic data to assess the degree of early digestive tract cancer.

[0111] The degree of early digestive tract cancer is the risk level corresponding to each cancer to be detected.

[0112] like Figure 2 As shown, the steps to assess the extent of early digestive tract cancer include:

[0113] Step d1: Preset the initial search interval [a0, b0]. In this embodiment, preferably a0=0, b0=100%; set the segmentation ratio h. In this embodiment, preferably h=0.618;

[0114] Step d2: According to the segmentation ratio h, the initial search interval [a0, b0] is divided into two new points ζ1 and ζ2, ζ1 = a0 + (1-h)(b0-a0), ζ2 = a0 + h(b0-a0); different digital labels are set for different cancers to be detected and marked as cancer labels; user inspection data and replacement feature data are marked as analysis data;

[0115] Step d3: Randomly select a cancer label that is not marked as a selected label and mark it as the evaluation label;

[0116] Step d4: respectively obtain the matching degrees corresponding to ζ1 and ζ2, mark the matching degree corresponding to ζ1 as the first matching degree, and mark the matching degree corresponding to ζ2 as the second matching degree;

[0117] Step d5: If the first matching degree is greater than or equal to the second matching degree, the minimum value of the updated search interval is the minimum value of the search interval before the update, and the maximum value is the point corresponding to the second matching degree; if the first matching degree is less than the second matching degree, the minimum value of the updated search interval is the point corresponding to the first matching degree, and the maximum value is the maximum value of the search interval before the update; illustratively, after the initial search interval [a0, b0] is updated through points ζ1 and ζ2, if the first matching degree is greater than or equal to the second matching degree, the updated search interval is [a0, ζ2], and if the first matching degree is less than the second matching degree, the updated search interval is [ζ1, b0];

[0118] Step d6: Determine whether to return to step d4; if so, return to step d4, if not, proceed to step d7;

[0119] Step d7: Determine whether all cancer tags are marked as selected tags. If so, obtain the degree of early digestive tract cancer; if not, return to step d3.

[0120] In the above step d4, the method for respectively obtaining the matching degree corresponding to ζ1 and ζ2 is: taking ζ1, analysis data and evaluation label as the first test data, and taking ζ2, analysis data and evaluation label as the second test data; inputting the first test data and the second test data into the trained matching analysis model respectively to predict the corresponding matching degree; the matching degree is the matching degree between the risk degree of the cancer to be detected corresponding to the analysis data and the evaluation label; the training process of the matching analysis model is consistent with the training process of the state judgment model, and both are deep neural network models.

[0121] In the above step d6, the method for determining whether to return to step d4 is: preset a width threshold, the width threshold is preset by a technician in this field according to actual conditions, and the interval width is obtained by subtracting the minimum value from the maximum value in the updated search interval; the interval width is compared with the width threshold; if the interval width is greater than or equal to the width threshold, the updated coefficient search area is re-divided into two new points ζ1 and ζ2 according to the segmentation ratio h, and the step d4 is returned; if the interval width is less than the width threshold, the average of the maximum and minimum values ​​in the updated coefficient search area is used as the evaluation label corresponding to the risk level of the cancer to be detected; and the evaluation label is marked as the selected label, and the step d7 is entered.

[0122] The suggestion generation module is used on the cloud platform to generate rehabilitation suggestions based on the degree of early digestive tract cancer and send them to the corresponding user end.

[0123] Rehabilitation suggestions include dietary suggestions and exercise suggestions; dietary suggestions include: for users with a low risk of the cancer to be detected, it is recommended to increase the intake of fiber-rich vegetables and fruits (such as green leafy vegetables, broccoli, apples, etc.); for users with a medium risk of the cancer to be detected, it is recommended to increase the intake of foods rich in antioxidants (such as blueberries and carrots) and foods rich in unsaturated fatty acids (such as salmon and nuts); exercise suggestions include: for users with a low risk of the cancer to be detected, it is recommended to maintain daily aerobic exercise habits and perform moderate-intensity exercise 3-4 times a week (such as brisk walking, jogging, cycling, etc.); for users with a medium risk of the cancer to be detected, light, low-impact exercise (such as walking, yoga, stretching, etc.) is recommended, each exercise lasts about 30 minutes, 4-5 times a week.

[0124] like Figure 3 As shown, an evolutionary algorithm is used to generate rehabilitation suggestions, and the steps include:

[0125] Evolutionary algorithms include genetic algorithms, clonal selection algorithms, etc. This embodiment takes the clonal selection algorithm as an example and uses an improved clonal selection algorithm to generate rehabilitation suggestions.

[0126] Step p1: Collect historical rehabilitation suggestions, set different digital labels for different rehabilitation suggestions in the historical rehabilitation suggestions, and mark them as suggestion labels;

[0127] Step p2: randomly generate an initial antibody group, where the antibodies in the initial antibody group correspond one-to-one to the recommended labels;

[0128] Step p3: determine the fitness function;

[0129] Step p4: Calculate the fitness of each antibody in the initial antibody group, and perform antibody screening to generate a memory antibody group;

[0130] Step p5: Calculate the corresponding clone size for each antibody in the memory antibody group, and clone it to generate a clone antibody group;

[0131] Step p6: Calculate the selection probability of each antibody in the cloned antibody group, and determine whether to select the antibody based on the selection probability to generate a selected antibody group;

[0132] Step p7: using an interpolation method to generate new antibodies for the antibodies in the selected antibody group to generate an interpolated antibody group;

[0133] Step p8: performing antibody recombination on the antibodies in the selected antibody group to generate a recombinant antibody group;

[0134] Step p9: merging the interpolated antibody group and the recombinant antibody group to generate a merged antibody group;

[0135] Step p10: clustering and screening the antibodies in the combined antibody group to generate a clustered antibody group;

[0136] Step p11: Determine whether the iteration is finished; if so, calculate the fitness of each antibody in the clustered antibody group, and obtain the rehabilitation advice corresponding to the recommended label corresponding to the antibody with the largest fitness; if not, regenerate the memory antibody group based on the clustered antibody group, and return to step p5.

[0137] In the above step p1, the historical rehabilitation suggestions are rehabilitation suggestions provided by the medical institution to the user at historical moments, and the historical rehabilitation suggestions are obtained from a relevant database of the medical institution.

[0138] In the above step p2, the initial antibody group X(0) = {X1, X2, X3, ... X q}, X q is the qth antibody, that is, the size of the initial antibody population is q; the range of antibodies is historical recovery recommendations; the number of iterations t of the initial antibody population is 0.

[0139] In the above step p3, the expression of the fitness function is: f=kx; wherein f is the fitness, and kx is the rehabilitation effect; the method for obtaining the rehabilitation effect is: taking the recommended label and analysis data corresponding to the antibody as evaluation data, inputting the evaluation data into the trained effect prediction model, and predicting the corresponding rehabilitation effect. The training process of the effect prediction model is consistent with the training process of the state judgment model, and both are deep neural network models; the rehabilitation effect is the degree of decrease in the degree of early gastrointestinal cancer after the user executes the rehabilitation suggestions.

[0140] In the above step p4, the method of antibody screening is: screening a antibodies with the largest fitness from the initial antibody group, q>a>1.

[0141] In the above step p5, the method for calculating the clone scale includes:

[0142]

[0143] Φ i =min(exp(||X i -X j ||));

[0144] In the formula, R i is the clone size of the i-th antibody, Int is the upward rounding function, q is the initial antibody population size, Φ i is the affinity of the ith antibody, f i is the fitness of the ith antibody, min is the minimum function, exp is the natural exponential function, ||X i -X j|| is the Euclidean distance between the i-th antibody and the j-th antibody, i≠j, j∈[1,q].

[0145] It should be noted that when calculating the Euclidean distance between the ith antibody and the jth antibody, normalization is performed so that 0≤||X i -X j ||≤1.

[0146] In the above step p6, the method for generating the selected antibody group comprises:

[0147] Calculate the fitness of each antibody in the cloned antibody group, and add them up in sequence to obtain the total antibody fitness, divide the fitness of each antibody by the total antibody fitness, and obtain the selection probability of each antibody; sort the selection probability of each antibody from large to small, and replace each selection probability in order according to the positive order with the sum of each selection probability and all selection probabilities before the corresponding selection probability; obtain all selection probabilities after replacement and mark them as replacement probabilities; the antibody corresponding to the replacement probability is consistent with the antibody corresponding to the selection probability before replacement; regard every two adjacent replacement probabilities as an analysis set according to the positive order; mark the replacement probability with the largest value in each analysis set as the first probability , mark the replacement probability with a smaller value as the second probability; construct the probability screening range corresponding to each analysis set according to the first probability and the second probability of each analysis set, wherein the first probability is the maximum value of the probability screening range, the second probability is the minimum value of the probability screening range, and the minimum value is an open interval, and the maximum value is a closed interval; each analysis set corresponds to the antibody corresponding to the second probability; wherein the minimum value of the probability screening range corresponding to the first antibody is 0, and the maximum value is the corresponding selection probability; randomly generate a random number between (0, 1], select the antibody corresponding to the probability screening range where the a random number is located, and generate a selected antibody group.

[0148] Exemplarily, the selection probabilities of the three antibodies are 0.6, 0.3, and 0.1, respectively; since there is no value before 0.6, 0.6 remains 0.6 after replacement; since 0.3 is preceded by 0.6, 0.3 is replaced by 0.3+0.6=0.9, and since 0.1 is preceded by 0.3 and 0.9, 0.1 is replaced by 0.1+0.3+0.6=1; therefore, the replacement probabilities are 0.5, 0.9, and 1, respectively; 0.5 and 0.9 are an analysis set, and the probability screening range corresponding to the second antibody is (0.5,0.9]; 0.9 and 1 are also an analysis set, and the probability screening range corresponding to the third antibody is (0.9,1]; the probability screening range corresponding to the first antibody is (0,0.5]; if the randomly generated random number is 0.8, since 0.8 is within (0.5,0.9], the second antibody is selected.

[0149] In the above step p7, the method for generating the interpolated antibody group includes:

[0150] Calculate the fitness of each antibody in the selected antibody group, and sort them from large to small to generate a sorting table; select half of the antibodies in the selected antibody group according to the positive order of the sorting table, and mark them as screening antibodies; use the interpolation method to generate new antibodies for every two adjacent screening antibodies; calculate the fitness corresponding to each new antibody, and add it to the sorting table, retain the antibodies corresponding to the first y fitnesses in the sorting table, and generate an interpolated antibody group, where y is the number of antibodies in the selected antibody group.

[0151] The expression of the new antibody is: X X =(1-α′)X′+α′X″; where X X is a new antibody, X′ and X″ are two adjacent screened antibodies, α′ is the interpolation coefficient, and α′ is a random number between [0, 1].

[0152] In the above step p8, the method for antibody recombination comprises:

[0153] Randomly select Y antibodies from the selected antibody group as parent antibodies for antibody recombination to generate progeny antibodies S c , until the number of unselected antibodies in the selected antibody group is less than Y, the antibody recombination is completed; in this embodiment, Y is preferably 3; the recombinant antibody group includes multiple progeny antibodies S c .

[0154] Progeny antibody S c The calculation methods include:

[0155]

[0156] In the formula, X′ Y is the Yth parent antibody, S Y is the Yth scaling factor, which is a randomly generated real number and not all Y scaling factors are 0.

[0157] In the above step p10, the method for generating a clustered antibody group includes:

[0158] Clustering algorithms (such as K-Means clustering, hierarchical clustering, DBSCAN, etc.) are used to cluster the antibodies in the merged antibody group to obtain u clusters, where u is an integer greater than 1; the antibodies in each cluster are added in turn, and then divided by the number of antibodies in the corresponding cluster to obtain the mean corresponding to each cluster, and marked as the cluster mean; the antibodies in each cluster are subtracted from the corresponding cluster mean in turn to obtain the antibody difference corresponding to each antibody; the antibody differences corresponding to the same cluster are compared to obtain the antibody difference with the smallest value corresponding to each cluster, and marked as the minimum difference; the antibody corresponding to the minimum difference corresponding to each cluster is used as the center of the corresponding cluster; and a clustered antibody group is generated according to the centers corresponding to the u clusters.

[0159] In the above step p11, the method for determining whether the iteration is finished includes:

[0160] Preset the iteration threshold, which is preset by those skilled in the art according to the algorithm accuracy;

[0161] Compare the number of iterations t with the iteration threshold;

[0162] If the number of iterations t is greater than or equal to the iteration threshold, the iteration ends;

[0163] If the number of iterations t is less than the iteration threshold, the maximum fitness in the combined antibody group is marked as the single maximum fitness; the single maximum fitness in the current iteration is subtracted from the single maximum fitness in the previous iteration and the absolute value is taken as the fitness change;

[0164] Preset variable threshold and quantity threshold, both of which are preset by those skilled in the art according to actual conditions; compare the fitness change with the variable threshold, if the fitness change is greater than or equal to the variable threshold, no smoothing instruction is generated; if the fitness change is less than the variable threshold, a smoothing instruction is generated; count the number of smoothing instructions generated and compare it with the quantity threshold;

[0165] If the number of smooth instructions is less than the number threshold, the iteration continues, and the number of iterations t=t+1. If u<a, au new antibodies are randomly generated to form a new memory antibody group with the clustered antibody group. If u≥a, a antibodies with the largest fitness are selected from the clustered antibody group to form a new memory antibody group. Among them, t=t+1 is a computer language, that is, the value of the number of iterations t is added by 1 and then assigned to the number of iterations t.

[0166] If the number of smoothing instructions is greater than or equal to the number threshold, the iteration ends.

[0167] It should be noted that the purpose of using evolutionary algorithms to generate rehabilitation suggestions is to:

[0168] 1. Optimize personalized matching of rehabilitation suggestions: By collecting historical rehabilitation suggestions, the evolutionary algorithm can generate customized rehabilitation plans based on the user's degree of early digestive tract cancer, and gradually optimize the applicability of personalized rehabilitation suggestions;

[0169] 2. Improve the accuracy and efficiency of suggestion generation: Adopt the iterative optimization process of evolutionary algorithm, through multiple rounds of antibody group screening, cloning and recombination, gradually optimize the generated suggestions to ensure the highest fitness of the final suggestions, thereby improving the accuracy and generation efficiency of rehabilitation suggestions;

[0170] 3. Enhance the exploration and diversity of the algorithm: The evolutionary algorithm includes operations such as antibody cloning, selection, interpolation and recombination, which improves the algorithm's ability to explore different rehabilitation suggestions, making the generated solutions diverse and comprehensive, and helping to discover potential high-quality rehabilitation suggestions;

[0171] 4. Adaptive adjustment of generated plans: The algorithm uses the fitness function and the probability of antibody selection to enable the generated rehabilitation plan to be adaptively adjusted. By calculating the fitness of antibodies and screening out suitable antibody groups, it can adaptively update the suggestions according to changes in user conditions, ensuring the dynamic adaptability of the suggestions.

[0172] 5. Reduce the impact of inappropriate suggestions: Through clustering and screening steps, rehabilitation suggestions with low fitness and that do not meet the needs can be effectively excluded, ensuring that the final output is the optimal suggestion suitable for the current user situation, reducing the interference of inappropriate suggestions on the rehabilitation process.

[0173] This embodiment collects user feature data and uses deep learning technology to accurately determine the state of the digestive tract, timely identify the user's potential digestive tract diseases, and encourage the user to conduct corresponding examinations in a timely manner; it adopts fuzzy logic control methods and search algorithms to achieve accurate estimation of the probability and risk of early digestive tract cancer, early detection and diagnosis of digestive tract cancer, and effectively avoid life safety risks caused by delayed treatment; at the same time, it adaptively generates rehabilitation suggestions through evolutionary algorithms, optimizes health management effects, and improves the screening and monitoring efficiency of high-risk groups, which helps to take treatment or protective measures in the early stage of digestive tract cancer, thereby improving the treatment effect, reducing the possibility of disease progression or complications, and thus improving the user's rehabilitation effect and long-term health status.

[0174] Example 2

[0175] See also Figure 4 As shown, the part not described in detail in this embodiment is described in Example 1, and a method for early warning of the risk of early cancer of the digestive tract based on data analysis is provided, which is applied in a cloud platform, and the method includes:

[0176] Collect user characteristic data;

[0177] Analyze user characteristic data to determine the user's digestive tract status;

[0178] Estimate the probability of early digestive tract cancer based on digestive tract status;

[0179] Collect user inspection data;

[0180] Combine user examination data and user characteristic data to assess the degree of early digestive tract cancer;

[0181] Generate rehabilitation recommendations based on the degree of early digestive tract cancer.

[0182] Furthermore, the digestive tract status includes normal, slightly abnormal, moderately abnormal and severely abnormal; if the digestive tract status is slightly abnormal, moderately abnormal or severely abnormal, the probability of early digestive tract cancer is estimated; if the digestive tract status is normal, the probability of early digestive tract cancer is not estimated; the probability of early digestive tract cancer is the probability corresponding to each type of digestive tract cancer;

[0183] The user characteristic data includes physiological data, life data, stress data and metabolic data; marking the non-numerical data in the user characteristic data as text data, setting different digital labels for different text data and marking them as data labels; replacing the text data in the user characteristic data with the corresponding data labels, and marking the replaced user characteristic data as replacement characteristic data;

[0184] The steps to estimate the probability of early digestive tract cancer include:

[0185] Step a1: Establish fuzzy sets, and divide each data in the replacement feature data into multiple fuzzy sets;

[0186] Step a2: convert each data in the replacement feature data into the membership degree of each corresponding fuzzy set through fuzzification technology; fuzzification is the process of converting an exact numerical value into the membership degree corresponding to the fuzzy set;

[0187] Step a3: define fuzzy rules;

[0188] Step a4: matching the fuzzified replacement feature data with the fuzzy rules, performing fuzzy reasoning, and obtaining fuzzy reasoning results, where the fuzzy reasoning results are the membership degrees corresponding to each type of digestive tract cancer;

[0189] Step a5: Use the fuzzy reasoning result as the probability of early digestive tract cancer.

[0190] Furthermore, the method for determining the digestive tract status of the user includes:

[0191] Input the replacement feature data into the trained state judgment model to predict the corresponding state label, where the state label is a digital label corresponding to the digestive tract state; and judge the user's digestive tract state based on the predicted state label;

[0192] The training process of the state judgment model includes:

[0193] Collect b groups of replacement feature data in advance, set corresponding state labels for the b groups of replacement feature data, where b is an integer greater than 1, and convert the replacement feature data and the corresponding state labels into a corresponding set of feature vectors;

[0194] Each group of feature vectors is used as the input of the state judgment model. The state judgment model takes a group of predicted state labels corresponding to each group of replacement feature data as output, and takes the actual state labels corresponding to each group of replacement feature data as prediction targets, where the actual state labels are pre-set state labels corresponding to the replacement feature data; minimizing the sum of prediction errors of all replacement feature data is used as the training target; the state judgment model is trained until the sum of prediction errors reaches convergence and the training is stopped; the state judgment model is a deep neural network model.

[0195] Furthermore, a probability threshold is preset, and the probability corresponding to each digestive tract cancer is compared with the probability threshold, and cancers with a probability greater than or equal to the probability threshold are marked as cancers to be detected, and cancers with a probability less than the probability threshold are not marked; the degree of early digestive tract cancer is the risk degree corresponding to each cancer to be detected;

[0196] The steps of evaluating the extent of early digestive tract cancer include:

[0197] Step d1: preset the initial search interval [a0, b0]; set the segmentation ratio h;

[0198] Step d2: According to the segmentation ratio h, the initial search interval [a0, b0] is divided into two new points ζ1 and ζ2, ζ1 = a0 + (1-h)(b0-a0), ζ2 = a0 + h(b0-a0); different digital labels are set for different cancers to be detected and marked as cancer labels; user inspection data and replacement feature data are marked as analysis data;

[0199] Step d3: Randomly select a cancer label that is not marked as a selected label and mark it as the evaluation label;

[0200] Step d4: respectively obtain the matching degrees corresponding to ζ1 and ζ2, mark the matching degree corresponding to ζ1 as the first matching degree, and mark the matching degree corresponding to ζ2 as the second matching degree;

[0201] Step d5: If the first matching degree is greater than or equal to the second matching degree, the minimum value of the updated search interval is the minimum value of the search interval before the update, and the maximum value is the point corresponding to the second matching degree; if the first matching degree is less than the second matching degree, the minimum value of the updated search interval is the point corresponding to the first matching degree, and the maximum value is the maximum value of the search interval before the update;

[0202] Step d6: Determine whether to return to step d4; if so, return to step d4, if not, proceed to step d7;

[0203] Step d7: Determine whether all cancer tags are marked as selected tags. If so, obtain the degree of early digestive tract cancer; if not, return to step d3.

[0204] Furthermore, in step d4, the method for respectively obtaining the matching degree corresponding to ζ1 and ζ2 is as follows: ζ1, analysis data and evaluation label are used as the first test data, and ζ2, analysis data and evaluation label are used as the second test data; the first test data and the second test data are respectively input into the trained matching analysis model to predict the corresponding matching degree; the matching degree is the matching degree between the risk degree of the cancer to be detected corresponding to the analysis data and the evaluation label; the training process of the matching analysis model is consistent with the training process of the state judgment model, and both are deep neural network models;

[0205] In step d6, the method for determining whether to return to step d4 is: presetting a width threshold, subtracting the minimum value from the maximum value in the updated search interval to obtain the interval width; comparing the interval width with the width threshold; if the interval width is greater than or equal to the width threshold, then re-dividing the updated coefficient search area into two new points ζ1 and ζ2 according to the segmentation ratio h, and returning to step d4; if the interval width is less than the width threshold, then taking the average of the maximum and minimum values ​​in the updated coefficient search area as the evaluation label corresponding to the risk level of the cancer to be detected; and marking the evaluation label as the selected label, and entering step d7.

[0206] Furthermore, an evolutionary algorithm is used to generate rehabilitation suggestions, the steps of which include:

[0207] Step p1: Collect historical rehabilitation suggestions, set different digital labels for different rehabilitation suggestions in the historical rehabilitation suggestions, and mark them as suggestion labels;

[0208] Step p2: randomly generate an initial antibody group, where the antibodies in the initial antibody group correspond one-to-one to the recommended labels;

[0209] Step p3: determine the fitness function;

[0210] Step p4: Calculate the fitness of each antibody in the initial antibody group, and perform antibody screening to generate a memory antibody group;

[0211] Step p5: Calculate the corresponding clone size for each antibody in the memory antibody group, and clone it to generate a clone antibody group;

[0212] Step p6: Calculate the selection probability of each antibody in the cloned antibody group, and determine whether to select the antibody based on the selection probability to generate a selected antibody group;

[0213] Step p7: using an interpolation method to generate new antibodies for the antibodies in the selected antibody group to generate an interpolated antibody group;

[0214] Step p8: performing antibody recombination on the antibodies in the selected antibody group to generate a recombinant antibody group;

[0215] Step p9: merging the interpolated antibody group and the recombinant antibody group to generate a merged antibody group;

[0216] Step p10: clustering and screening the antibodies in the combined antibody group to generate a clustered antibody group;

[0217] Step p11: Determine whether the iteration is finished; if so, calculate the fitness of each antibody in the clustered antibody group, and obtain the rehabilitation advice corresponding to the recommended label corresponding to the antibody with the largest fitness; if not, regenerate the memory antibody group based on the clustered antibody group, and return to step p5.

[0218] Furthermore, in step p2, the initial antibody group X(0) = {X1, X2, X3, ... X q}, X q is the qth antibody, that is, the size of the initial antibody population is q; the range of antibodies is historical recovery suggestions; the number of iterations t of the initial antibody population is 0;

[0219] In step p3, the fitness function is expressed as: f=kx; wherein f is fitness, and kx is rehabilitation effect; the method for obtaining the rehabilitation effect is: taking the recommended label and analysis data corresponding to the antibody as evaluation data, inputting the evaluation data into the trained effect prediction model, and predicting the corresponding rehabilitation effect, the training process of the effect prediction model is consistent with the training process of the state judgment model, and both are deep neural network models; the rehabilitation effect is the degree of decrease in the degree of early digestive tract cancer after the user executes the rehabilitation suggestion;

[0220] In step p4, the method of antibody screening is: screening a antibodies with the largest fitness from the initial antibody group, q>a>1;

[0221] In step p5, the method for calculating the clone scale includes:

[0222]

[0223] Φ i =min(exp(||X i -X j ||));

[0224] In the formula, R i is the clone size of the i-th antibody, Int is the upward rounding function, q is the initial antibody population size, Φ i is the affinity of the ith antibody, f i is the fitness of the ith antibody, min is the minimum function, exp is the natural exponential function, ||X i -X j|| is the Euclidean distance between the i-th antibody and the j-th antibody, i≠j, j∈[1,q].

[0225] Furthermore, in step p6, the method for generating the selected antibody group includes:

[0226] Calculate the fitness of each antibody in the cloned antibody group, and add them up in sequence to obtain the total antibody fitness, divide the fitness of each antibody by the total antibody fitness, and obtain the selection probability of each antibody; sort the selection probability of each antibody from large to small, and replace each selection probability in order according to the positive order with the sum of each selection probability and all selection probabilities before the corresponding selection probability; obtain all selection probabilities after replacement and mark them as replacement probabilities; the antibody corresponding to the replacement probability is consistent with the antibody corresponding to the selection probability before replacement; regard every two adjacent replacement probabilities as an analysis set according to the positive order; mark the replacement probability with the largest value in each analysis set as the first probability , mark the replacement probability with a smaller value as the second probability; construct the probability screening range corresponding to each analysis set according to the first probability and the second probability of each analysis set, wherein the first probability is the maximum value of the probability screening range, the second probability is the minimum value of the probability screening range, and the minimum value is an open interval, and the maximum value is a closed interval; each analysis set corresponds to the antibody corresponding to the second probability; wherein the minimum value of the probability screening range corresponding to the first antibody is 0, and the maximum value is the corresponding selection probability; randomly generate a random number between (0, 1], select the antibody corresponding to the probability screening range where the a random number is located, and generate a selected antibody group;

[0227] In step p7, the method for generating an interpolated antibody group includes:

[0228] Calculate the fitness of each antibody in the selected antibody group, and sort them from large to small to generate a sorting table; select half of the antibodies in the selected antibody group according to the positive order of the sorting table, and mark them as screening antibodies; use the interpolation method to generate new antibodies for every two adjacent screening antibodies; calculate the fitness corresponding to each new antibody, and add it to the sorting table, retain the antibodies corresponding to the first y fitnesses in the sorting table, and generate an interpolated antibody group, where y is the number of antibodies in the selected antibody group;

[0229] The expression of the new antibody is: X X =(1-α′)X′+α′X″; where X X is a new antibody, X′ and X″ are two adjacent screened antibodies, α′ is the interpolation coefficient, and α′ is a random number between [0, 1].

[0230] Furthermore, in step p8, the method for antibody recombination comprises:

[0231] Randomly select Y antibodies from the selected antibody group as parent antibodies for antibody recombination to generate progeny antibodies S c , until the number of unselected antibodies in the selected antibody group is less than Y, the antibody recombination is completed; the recombinant antibody group includes multiple progeny antibodies S c ;

[0232] In step p10, the method for generating a clustered antibody group includes: clustering the antibodies in the merged antibody group using a clustering algorithm to obtain u clusters, where u is an integer greater than 1; adding the antibodies in each cluster in turn, and then dividing by the number of antibodies in the corresponding cluster to obtain the mean corresponding to each cluster, and marking it as the cluster mean; subtracting the corresponding cluster mean from the antibodies in each cluster in turn to obtain the antibody difference corresponding to each antibody; comparing the antibody differences corresponding to the same cluster, obtaining the antibody difference with the minimum value corresponding to each cluster, and marking it as the minimum difference; taking the antibody corresponding to the minimum difference corresponding to each cluster as the center of the corresponding cluster; generating a clustered antibody group according to the centers corresponding to the u clusters;

[0233] In step p11, the method for determining whether the iteration is finished includes:

[0234] An iteration threshold is preset, and the number of iterations t is compared with the iteration threshold; if the number of iterations t is greater than or equal to the iteration threshold, the iteration ends; if the number of iterations t is less than the iteration threshold, the maximum fitness in the combined antibody group is marked as the single maximum fitness; the single maximum fitness in the current iteration process is subtracted from the single maximum fitness in the previous iteration process and the absolute value is taken as the fitness change;

[0235] Preset variable threshold and quantity threshold; compare the fitness change with the variable threshold, if the fitness change is greater than or equal to the variable threshold, do not generate a smooth instruction; if the fitness change is less than the variable threshold, generate a smooth instruction; count the number of smooth instructions generated and compare with the quantity threshold; if the number of smooth instructions is less than the quantity threshold, the iteration continues, and the number of iterations t=t+1, if u<a, randomly generate au new antibodies and form a new memory antibody group with the clustered antibody group, if u≥a, select a antibodies with the largest fitness from the clustered antibody group to form a new memory antibody group; if the number of smooth instructions is greater than or equal to the quantity threshold, the iteration ends.

[0236] Example 3

[0237] The present application also provides an electronic device. The electronic device may include one or more processors and one or more memories. The memories store computer-readable codes, and when the computer-readable codes are executed by the one or more processors, the method for early warning of the risk of early gastrointestinal cancer based on data analysis as described above may be executed.

[0238] The method or system according to the implementation mode of the present application can also be implemented with the help of the architecture of the electronic device shown in the present application. The electronic device may include a bus, one or more CPUs, ROM, RAM, a communication port connected to a network, input / output, a hard disk, etc. A storage device in an electronic device, such as a ROM or a hard disk, can store a method for warning the risk of early gastrointestinal cancer based on data analysis provided in the present application. Furthermore, the electronic device may also include a user interface. Of course, the architecture shown in the present application is only exemplary. When implementing different devices, one or more components in the electronic device shown in the present application may be omitted according to actual needs.

[0239] Example 4

[0240] One embodiment of the present application discloses a computer-readable storage medium. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are executed by the processor, a method for early warning of the risk of early cancer of the digestive tract based on data analysis according to an embodiment of the present application described with reference to the above figures can be executed. The storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0241] In addition, according to the embodiments of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided in the present application, for example: a method for early warning of the risk of early cancer of the digestive tract based on data analysis. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0242] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0243] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for early warning of the risk of early digestive tract cancer based on data analysis, characterized in that: Applied in cloud platforms, including: Collect user characteristic data; Analyze user characteristic data to determine the user's digestive tract status; Estimate the probability of early digestive tract cancer based on digestive tract status; Collect user inspection data; Combine user examination data and user characteristic data to assess the degree of early digestive tract cancer; Generate rehabilitation recommendations based on the degree of early digestive tract cancer.

2. The method for early warning of the risk of early gastrointestinal cancer based on data analysis according to claim 1, characterized in that: The digestive tract status includes normal, slightly abnormal, moderately abnormal and severely abnormal; if the digestive tract status is slightly abnormal, moderately abnormal or severely abnormal, the probability of early digestive tract cancer is estimated; if the digestive tract status is normal, the probability of early digestive tract cancer is not estimated; the probability of early digestive tract cancer is the probability corresponding to each type of digestive tract cancer; The user characteristic data includes physiological data, life data, stress data and metabolic data; marking the non-numerical data in the user characteristic data as text data, setting different digital labels for different text data and marking them as data labels; replacing the text data in the user characteristic data with the corresponding data labels, and marking the replaced user characteristic data as replacement characteristic data; The steps to estimate the probability of early digestive tract cancer include: Step a1: Establish fuzzy sets, and divide each data in the replacement feature data into multiple fuzzy sets; Step a2: convert each data in the replacement feature data into the membership degree of each corresponding fuzzy set through fuzzification technology; fuzzification is the process of converting an exact numerical value into the membership degree corresponding to the fuzzy set; Step a3: define fuzzy rules; Step a4: matching the fuzzified replacement feature data with the fuzzy rules, performing fuzzy reasoning, and obtaining fuzzy reasoning results, where the fuzzy reasoning results are the membership degrees corresponding to each type of digestive tract cancer; Step a5: Use the fuzzy reasoning result as the probability of early digestive tract cancer.

3. The method for early warning of the risk of early gastrointestinal cancer based on data analysis according to claim 2, characterized in that: The method for determining the digestive tract status of the user includes: Input the replacement feature data into the trained state judgment model to predict the corresponding state label, where the state label is a digital label corresponding to the digestive tract state; and judge the user's digestive tract state based on the predicted state label; The training process of the state judgment model includes: Collect b groups of replacement feature data in advance, set corresponding state labels for the b groups of replacement feature data, where b is an integer greater than 1, and convert the replacement feature data and the corresponding state labels into a corresponding set of feature vectors; Each group of feature vectors is used as the input of the state judgment model. The state judgment model takes a group of predicted state labels corresponding to each group of replacement feature data as output, and takes the actual state labels corresponding to each group of replacement feature data as prediction targets, where the actual state labels are pre-set state labels corresponding to the replacement feature data; minimizing the sum of prediction errors of all replacement feature data is used as the training target; the state judgment model is trained until the sum of prediction errors reaches convergence and the training is stopped; the state judgment model is a deep neural network model.

4. The method for early warning of the risk of early digestive tract cancer based on data analysis according to claim 3, characterized in that: A probability threshold is preset, and the probability corresponding to each digestive tract cancer is compared with the probability threshold, and cancers with a probability greater than or equal to the probability threshold are marked as cancers to be detected, and cancers with a probability less than the probability threshold are not marked; the degree of early digestive tract cancer is the risk degree corresponding to each cancer to be detected; The steps of evaluating the extent of early digestive tract cancer include: Step d1: preset the initial search interval [a0, b0]; set the segmentation ratio h; Step d2: According to the segmentation ratio h, the initial search interval [a0, b0] is divided into two new points ζ1 and ζ2, ζ1 = a0 + (1-h)(b0-a0), ζ2 = a0 + h(b0-a0); different digital labels are set for different cancers to be detected and marked as cancer labels; user inspection data and replacement feature data are marked as analysis data; Step d3: Randomly select a cancer label that is not marked as a selected label and mark it as the evaluation label; Step d4: respectively obtain the matching degrees corresponding to ζ1 and ζ2, mark the matching degree corresponding to ζ1 as the first matching degree, and mark the matching degree corresponding to ζ2 as the second matching degree; Step d5: If the first matching degree is greater than or equal to the second matching degree, the minimum value of the updated search interval is the minimum value of the search interval before the update, and the maximum value is the point corresponding to the second matching degree; if the first matching degree is less than the second matching degree, the minimum value of the updated search interval is the point corresponding to the first matching degree, and the maximum value is the maximum value of the search interval before the update; Step d6: Determine whether to return to step d4; if so, return to step d4, if not, proceed to step d7; Step d7: Determine whether all cancer tags are marked as selected tags. If so, obtain the degree of early digestive tract cancer; if not, return to step d3.

5. The method for early warning of the risk of early gastrointestinal cancer based on data analysis according to claim 4, characterized in that: In the step d4, the method for respectively obtaining the matching degree corresponding to ζ1 and ζ2 is as follows: ζ1, analysis data and evaluation label are used as the first test data, and ζ2, analysis data and evaluation label are used as the second test data; the first test data and the second test data are respectively input into the trained matching analysis model to predict the corresponding matching degree; the matching degree is the matching degree between the risk degree of the cancer to be detected corresponding to the analysis data and the evaluation label; the training process of the matching analysis model is consistent with the training process of the state judgment model, and both are deep neural network models; In step d6, the method for determining whether to return to step d4 is: presetting a width threshold, subtracting the minimum value from the maximum value in the updated search interval to obtain the interval width; comparing the interval width with the width threshold; If the interval width is greater than or equal to the width threshold, the updated coefficient search area is re-divided into two new points ζ1 and ζ2 according to the segmentation ratio h, and the process returns to step d4; if the interval width is less than the width threshold, the average of the maximum and minimum values ​​in the updated coefficient search area is used as the evaluation label corresponding to the risk level of the cancer to be detected; and the evaluation label is marked as the selected label, and the process goes to step d7.

6. The method for early warning of the risk of early gastrointestinal cancer based on data analysis according to claim 5, characterized in that: The evolutionary algorithm is used to generate rehabilitation suggestions, and the steps include: Step p1: Collect historical rehabilitation suggestions, set different digital labels for different rehabilitation suggestions in the historical rehabilitation suggestions, and mark them as suggestion labels; Step p2: randomly generate an initial antibody group, where the antibodies in the initial antibody group correspond one-to-one to the recommended labels; Step p3: determine the fitness function; Step p4: Calculate the fitness of each antibody in the initial antibody group, and perform antibody screening to generate a memory antibody group; Step p5: Calculate the corresponding clone size for each antibody in the memory antibody group, and clone it to generate a clone antibody group; Step p6: Calculate the selection probability of each antibody in the cloned antibody group, and determine whether to select the antibody based on the selection probability to generate a selected antibody group; Step p7: using an interpolation method to generate new antibodies for the antibodies in the selected antibody group to generate an interpolated antibody group; Step p8: performing antibody recombination on the antibodies in the selected antibody group to generate a recombinant antibody group; Step p9: merging the interpolated antibody group and the recombinant antibody group to generate a merged antibody group; Step p10: clustering and screening the antibodies in the combined antibody group to generate a clustered antibody group; Step p11: Determine whether the iteration is finished; if so, calculate the fitness of each antibody in the clustered antibody group, and obtain the rehabilitation advice corresponding to the recommended label corresponding to the antibody with the largest fitness; if not, regenerate the memory antibody group based on the clustered antibody group, and return to step p5.

7. The method for early warning of the risk of early digestive tract cancer based on data analysis according to claim 6, characterized in that: In step p2, the initial antibody group X(0) = {X1, X2, X3, ... X q }, X q is the qth antibody, that is, the size of the initial antibody population is q; the range of antibodies is historical recovery suggestions; the number of iterations t of the initial antibody population is 0; In step p3, the fitness function is expressed as: f=kx; wherein f is fitness, and kx is rehabilitation effect; the method for obtaining the rehabilitation effect is: taking the recommended label and analysis data corresponding to the antibody as evaluation data, inputting the evaluation data into the trained effect prediction model, and predicting the corresponding rehabilitation effect, the training process of the effect prediction model is consistent with the training process of the state judgment model, and both are deep neural network models; the rehabilitation effect is the degree of decrease in the degree of early digestive tract cancer after the user executes the rehabilitation suggestion; In step p4, the method of antibody screening is: screening a antibodies with the largest fitness from the initial antibody group, q>a>1; In step p5, the method for calculating the clone scale includes: F i =min(exp(||X i -X j ||)); In the formula, R i is the clone size of the i-th antibody, Int is the upward rounding function, q is the initial antibody population size, Φ i is the affinity of the ith antibody, f i is the fitness of the ith antibody, min is the minimum function, exp is the natural exponential function, ||X i -X j || is the Euclidean distance between the i-th antibody and the j-th antibody, i≠j, j∈[1,q].

8. The method for early warning of the risk of early digestive tract cancer based on data analysis according to claim 7, characterized in that: In step p6, the method for generating the selected antibody group comprises: Calculate the fitness of each antibody in the cloned antibody group, and add them up in sequence to obtain the total antibody fitness, divide the fitness of each antibody by the total antibody fitness, and obtain the selection probability of each antibody; sort the selection probability of each antibody from large to small, and replace each selection probability in order according to the positive order with the sum of each selection probability and all selection probabilities before the corresponding selection probability; obtain all selection probabilities after replacement and mark them as replacement probabilities; the antibody corresponding to the replacement probability is consistent with the antibody corresponding to the selection probability before replacement; regard every two adjacent replacement probabilities as an analysis set according to the positive order; mark the replacement probability with the largest value in each analysis set as the first probability , mark the replacement probability with a smaller value as the second probability; construct the probability screening range corresponding to each analysis set according to the first probability and the second probability of each analysis set, wherein the first probability is the maximum value of the probability screening range, the second probability is the minimum value of the probability screening range, and the minimum value is an open interval, and the maximum value is a closed interval; each analysis set corresponds to the antibody corresponding to the second probability; wherein the minimum value of the probability screening range corresponding to the first antibody is 0, and the maximum value is the corresponding selection probability; randomly generate a random number between (0, 1], select the antibody corresponding to the probability screening range where the a random number is located, and generate a selected antibody group; In step p7, the method for generating the interpolated antibody group includes: Calculate the fitness of each antibody in the selected antibody group, and sort them from large to small to generate a sorting table; select half of the antibodies in the selected antibody group according to the positive order of the sorting table, and mark them as screening antibodies; use the interpolation method to generate new antibodies for every two adjacent screening antibodies; calculate the fitness corresponding to each new antibody, and add it to the sorting table, retain the antibodies corresponding to the first y fitnesses in the sorting table, and generate an interpolated antibody group, where y is the number of antibodies in the selected antibody group; The expression of the new antibody is: X X =(1-α′)X′+α′X″; where X X is a new antibody, X′ and X″ are two adjacent screened antibodies, α′ is the interpolation coefficient, and α′ is a random number between [0, 1].

9. The method for early warning of the risk of early digestive tract cancer based on data analysis according to claim 8, characterized in that: In step p8, the method for antibody recombination comprises: Randomly select Y antibodies from the selected antibody group as parent antibodies for antibody recombination to generate progeny antibodies S c , until the number of unselected antibodies in the selected antibody group is less than Y, the antibody recombination is completed; the recombinant antibody group includes multiple progeny antibodies S c ; In step p10, the method for generating a clustered antibody group includes: clustering the antibodies in the merged antibody group using a clustering algorithm to obtain u clusters, where u is an integer greater than 1; adding the antibodies in each cluster in turn, and then dividing by the number of antibodies in the corresponding cluster to obtain the mean corresponding to each cluster, and marking it as the cluster mean; subtracting the corresponding cluster mean from the antibodies in each cluster in turn to obtain the antibody difference corresponding to each antibody; comparing the antibody differences corresponding to the same cluster, obtaining the antibody difference with the minimum value corresponding to each cluster, and marking it as the minimum difference; taking the antibody corresponding to the minimum difference corresponding to each cluster as the center of the corresponding cluster; generating a clustered antibody group according to the centers corresponding to the u clusters; In step p11, the method for determining whether the iteration is finished includes: An iteration threshold is preset, and the number of iterations t is compared with the iteration threshold; if the number of iterations t is greater than or equal to the iteration threshold, the iteration ends; if the number of iterations t is less than the iteration threshold, the maximum fitness in the combined antibody group is marked as the single maximum fitness; the single maximum fitness in the current iteration process is subtracted from the single maximum fitness in the previous iteration process and the absolute value is taken as the fitness change; Preset variable threshold and quantity threshold; compare the fitness change with the variable threshold, if the fitness change is greater than or equal to the variable threshold, do not generate a smooth instruction; if the fitness change is less than the variable threshold, generate a smooth instruction; count the number of smooth instructions generated and compare with the quantity threshold; if the number of smooth instructions is less than the quantity threshold, the iteration continues, and the number of iterations t=t+1, if u<a, randomly generate au new antibodies and form a new memory antibody group with the clustered antibody group, if u≥a, select a antibodies with the largest fitness from the clustered antibody group to form a new memory antibody group; if the number of smooth instructions is greater than or equal to the quantity threshold, the iteration ends.

10. A digestive tract early cancer risk warning system based on data analysis, implementing a digestive tract early cancer risk warning method based on data analysis as claimed in any one of claims 1 to 9, characterized in that: Applied in cloud platforms, including: A data collection module, used to collect user characteristic data; The data analysis module is used to analyze the user's characteristic data and determine the user's digestive tract status; The probability estimation module estimates the probability of early digestive tract cancer based on the digestive tract status; A data acquisition module, used to collect user inspection data; The risk assessment module is used to combine user examination data and user characteristic data to assess the degree of early digestive tract cancer; The suggestion generation module generates rehabilitation suggestions based on the degree of early digestive tract cancer.

Images